What kind of deception is AI?

The development of technology is a double-edged sword. While bringing people a convenient experience, it will also be targeted by many lawless elements, resulting in various new fraud methods, among which AI fraud is one. In order not to be harvested by scammers, it is an essential lesson to know what capabilities AI has in advance.

Contemporary artificial intelligence deception

0 1AI robot phone I think everyone has experienced such a scene: when the official call is connected, the other end of the voice will naturally greet you. After the pause, if you politely respond to "who", the robot on the other side will start to greet you and encourage you to handle business. What's more, AI robots will use false excuses to ask for money or personal information.

Robot phone is a wide application of AI speech synthesis technology. Similar AI robot mobile phone fraud has also been cracked by public security organs in recent years. However, there is no other way to effectively stop illegal robot calls except that users actively mark "fraud numbers".

According to The Wall Street Journal, the CEO of an energy company in Britain received a phone call from his parent company, and the person calling himself "CEO" called for money three times a day. He was asked to remit money to Hungarian suppliers, claiming that in order to stabilize the cooperative relationship between the two parties, he finally succeeded in defrauding 220,000 euros.

It is worth noting that the caller's English has a little German accent, which is the same as the boss's melody and exactly the same as the boss of the parent company. At least judging from the voice, it must be the boss himself.

This is artificial intelligence voice cloning technology. In addition to using neural networks to model and imitate the original audio, AI can also analyze my social network and quickly master the personalized way of speaking, the relationship with people around me, hobbies and so on. To imitate your natural communication with the people around you. Even special accents or pure dialects can be synthesized.

03AI's face-changing effect is easier to gain the trust of the other party. Liars can use AI technology to change their faces and pretend to be anyone, and then confirm the information through video. The face-changing App“ZAO ",which was very popular some time ago, can change one person's face into another person's figure. As long as you find enough material, you can completely "change your face" and everyone can get started.

On the live broadcast platform, some people pretend to be Roy, but they can still fool fans. This is the application of changing face and voice.

Screening the deceived through AI technology. Liars don't cast nets aimlessly, but aim at specific objects with ulterior motives. They will screen the crowd according to the deception to be implemented, so as to select the target object.

Give a chestnut: when conducting financial fraud, small partners who often collect investment information will become their potential targets.

Using AI technology, coupled with the interpretation of routines and scripts, can such fraudulent means be hold by friends?

AI fraud, how to effectively prevent it?

How to effectively prevent and prevent more people from being deceived may require the joint efforts of policies, technologies and individuals.

1. Technology With the intensification of AI fraud, many new technical methods have been applied to prevent AI fraud. For example, Symantec, a security company, recently proposed a method of using blockchain technology and voice over IP (VoIP) calls to identify callers, thus reducing the number of fraudulent calls from superiors.

Without knowing the consequences of technology diffusion, reasonably releasing technological achievements has become the choice of some technology enterprises. For example, GPT-2, an unsupervised language model with higher performance launched by OpenAI some time ago, only released a simplified version, without releasing data sets, training codes and model weights.

2. Policy In addition to the wits and wits between technicians and hackers, the industry has also begun to set up a restricted area for the abuse of AI from the level of rule construction.

In recent years, both FTC and FCC have stepped up their supervision actions against illegal robot phones, and issued notices to eight small telecom operators and Internet communication companies, urging them to trace and close the suspected fraudulent phone sources.

In the report "Artificial Intelligence and National Security" issued by the American Security Center, artificial intelligence forgery technology is also clearly listed as a key technology that threatens national security.

China has also begun to deal with the potential security risks of artificial intelligence through policy management and technical restrictions.

3. Personal protection of personal information. In order to prevent scammers from using AI technology to process personal information and thus cheat, friends should guard against all kinds of information leakage, whether on the Internet or social software, and try to avoid exposing their own information too much.

Certification. When transactions such as transfer are involved, if the identity of the other party cannot be confirmed, the arrival time can be set to "24 hours" and the processing time can be reserved. Even if the other party cheats by using AI technology, they can further confirm their identity by asking questions. I suggest you transfer money to the other bank account and avoid transferring money through social software such as WeChat. On the one hand, it helps to verify the identity of the other party, on the other hand, it also helps to follow up the transfer information.

Prompt each other and take precautions together. Remind and warn relatives and friends around us, improve safety awareness and ability to deal with high-tech fraud, and jointly prevent being deceived, especially the elderly at home. In the face of all kinds of online fraud, we must be alert to three points!